Inhalt und Ziele des Forschungsprojekts
Die Tiefe Hirnstimulation ist zentraler Bestandteil der Therapie der fortgeschrittenen Parkinsonerkrankung. Für das Sprechen ist das fein koordinierte Zusammenspiel zahlreicher Muskelgruppen erforderlich. Die Parkinsonkrankheit führt als Bewegungsstörung früh zu einer Beeinträchtigung des Sprechens. Die Tiefe Hirnstimulation kann die parkinsonbedingte Sprechstörung zwar einerseits verbessern, sie kann andererseits aber auch selbst zu Sprechstörungen führen (z.B. undeutliche Aussprache oder gepresste Stimme). Das menschliche Ohr ist nicht genau genug, zu unterscheiden, ob eine Sprechstörung durch die Parkinsonerkrankung, oder durch die Tiefe Hirnstimulation verursacht ist. In der ersten Phase unseres Projektes wird daher eine Spracherkennungssoftware mittels künstlicher Intelligenz trainiert, die Einflüsse von Parkinsonerkrankung und Tiefer Hirnstimulation auf das Sprechen zu unterscheiden. In der zweiten Phase werden moderne Bildgebungsverfahren eingesetzt, um die Hirnstrukturen zu identifizieren, die an der Entstehung von Sprechstörungen durch Tiefen Hirnstimulation beteiligt sind. In der dritten Phase wird in einer klinischen Studie untersucht, ob mit Hilfe der neuen Software und den Kenntnissen, welche Hirnstrukturen für Sprechstörungen verantwortlich sind, die Tiefe Hirnstimulation dahingehend optimiert werden kann, dass Sprechstörungen nicht mehr auftreten.
Wissenschaftlicher und gesellschaftlicher Kontext des Forschungsprojekts
Ziel dieses Projekts ist es, Sprechstörung bei Parkinsonpatienten unter Therapie mit Tiefer Hirnstimulation zu verbessern. Weiter setzt dieses Projekt durch den Einsatz von künstlicher Intelligenz einen Grundstein zur automatisierten Messung von Parkinsonsymptomen sowie zur automatisierten Anpassung der Therapie, und leistet damit einen Beitrag zur Digitalisierung in der Medizin.
REducing SPEECH-related side-effects of deep brain stimulation in Parkinson's disease via automated speech analysis (RESPEECH-PD)Background. Deep brain stimulation of the subthalamic nucleus (STN-DBS) is an effective treatment of L-dopa sensitive motor symptoms of Parkinson’s disease (PD), but its effects on speech are equivocal. Speech is the most complex learned motor program and it is of uppermost importance for social interaction. Although some aspects of speech might improve with STN-DBS, stimulation-induced dysarthria represents the most common side effect, with a prevalence of up to 90%. Worsening of speech can neutralize the motor benefits of STN-DBS in terms of overall benefit in quality of life. The pathophysiology behind STN-DBS-induced dysarthria includes current diffusion to myelinated fiber tracts surrounding the STN, mainly the pyramidal tract, located at the lateral border of the sensorimotor STN, leading to dysarthria. STN-DBS-induced dysarthria has to be distinguished from parkinsonian dysarthria. Unfortunately, the human ear has very limited capacities in the qualitative and quantitative evaluation of speech; thus improvement of STN-DBS induced dysarthria is still an unmet need. New methods of automated speech assesment based on acoustic analysis and machine-learning algorithms have quantified subtle speech alterations and distinguished different types of dysarthria. These new tools are outperforming the capacity of speech analysis of the most skilled clinicians. Modern directional DBS-systems allow for current steering, bearing the potential of reducing or completely avoiding stimulation-induced dysarthria. To leverage steering, detailed stimulation maps and improved dysarthria analysis are required that indicate regions of therapeutic effect and side effects. Aims and methods. The applicants unite several experts with complementary and proven high expertise in postoperative management of STN-DBS for PD, automated speech analysis in PD, imaging of precise electrode position and tractography, as well as modelling of current diffusion, which are fundamental prerequisites for achieving the following aims:Aim 1.To identify the most sensitive and specific speech variables for STN-DBS-related improvement of parkinsonian dysarthria and STN-DBS-induced speech-related side-effects, by application of an automated speech analysis technique, comparing stimulation off and on states, as well as increasing stimulation amplitudes. Aim 2.Mapping of stimulation related speech changes: to investigate the anatomical and pathophysiological substrate of STN-DBS induced changes in speech production. Dysarthria severity, as measured by automated speech analysis, volume of tissue activated (VTA), and tractography of fiber tracts adjacent to the STN, will be set into context to determine a sweet spot for dysarthria reduction and to reveal speech parameters that correlate with current diffusion to the pyramidal tract. Aim 3.To investigate whether an automated speech analysis technique reliably detects improvements of STN-DBS-induced dysarthria in a longitudinal interventional cross-over study and to compare sensitivity for dysarthria reduction between automated, perceptive, and subjective speech analysis. Expected results. With successful completion of the presented protocol, we will be able to reliably detect the footprint of STN-DBS in speech and to identify the anatomical substrates for both improvement and deterioration of speech in PD, laying the foundation for better conventional and closed-loop postoperative management of STN-DBS in PD.Expected value. Systematic implementation of automated speech analysis would facilitate not only time-consuming postoperative management but also improve overall outcomes of STN-DBS in PD patients treated with STN-DBS. This would be the first use-inspired step towards fully automatized closed-loop DBS based on speech as a biomarker, allowing to standardize and improve outcomes of DBS across centers worldwide in the near future.
Last updated:18.07.2023
